Even decades after the advent of automatic seismic analysis routines and further development, the gold standard for passive seismic processing has long been manually picked phase arrivals. Only recently seismic data analysis performed with advanced automated waveform based methods and/or Deep Learning approaches allowed to obtain results comparable or, in some cases superior, to the manual ones. The massive growth of passive seismic data sets from hyper-dense surface array of nodal instruments such as the ones foreseen in WP1, massive arrays of geophones with thousands of sensors and distributed acoustic sensing (WP1) makes this paradigm shift even more urgent. Template matching techniques yield catalogs that are usually more complete by one or two orders of magnitude. With the many smaller events detected it becomes a greater challenge to associate the detected arrivals to the correct event. This is particularly a problem for EGS simulations, where we deliberately create thousands of small microseismic events. Here, we have to frequently deal with the case of overlapping wave trains.
Furthermore, information is extracted from the ambient noise using beamforming and/or interferometric methods. Analysis of the ambient seismic noise may enable the characterisation of structures in the subsurface and subsequently increase the safety of EGS applications. The combination of ambient seismic noise methods complements waveform-based methods that locate and characterise seismic events. Workflows will be developed to integrate the developed technologies of WP1 into the innovative methods developed in WP2 and applied to large heterogeneous datasets. A set of techniques for real-time forecasting of induced seismicity and risk assessment will be shared with WP 3 - 5.
Lead: IEG
Participants: ETH, LBNL, GES, EOST, DIAS
Contact: Erik Saenger & Claudia Finger
Task 2.1: Exploiting machine learning techniques for microseismicity characterization
Task 2.2: Integrating DAS and seismometer chains in real-time seismic and deformation
analysis workflows for monitoring and imaging
Task 2.3: Develop new imaging approaches to borehole-based seismic imaging, targeted at
detecting major fracture zones and faults near a borehole
Task 2.4: Apply beamforming methods to estimate anisotropy and derive information about
seismic velocities
Task 2.5: Time-Reverse Imaging for locating and characterizing micro-seismicity
D 2.1: Deep learning methods for microseismicity characterization
D 2.2: Comparison of different techniques for induced seismicity monitoring
D 2.3: Integrating DAS and geophone chain in real-time monitoring systems
D 2.4: Using DAS and geophone chain for imaging (fault detection)
A list of all deliverables for DEEP is available here.